Skip to content

README.md

Latest commit

 

History

History
44 lines (29 loc) · 1.68 KB

README.md

File metadata and controls

44 lines (29 loc) · 1.68 KB

ByteTorch

ByteTorch is a small deep learning library written in Python, inspired by PyTorch. It's a personal project for learning and experimenting with neural networks from the inside. The code is simple and open - feel free to read, use, or break things as you like.

What you'll find here

  • Custom autograd engine - a minimal implementation of automatic differentiation and backpropagation
  • Clear and well-organized code - designed to be easy to read and modify
  • Essential neural network layers - including Linear, activation functions, Dropout, and BatchNorm
  • Optimizers - SGD, Adam, and a straightforward way to add your own
  • Example Jupyter notebooks - practical demonstrations for regression and classification

Why ByteTorch?

ByteTorch is designed for those who want to better understand the inner workings of deep learning frameworks. It can serve as a learning resource, a base for experimenting with custom layers or optimizers, or a lightweight tool for small projects and educational purposes.

Getting Started

  1. Clone the repository:

    git clone https://github.com/ffenix7/ByteTorch.git
cd ByteTorch

  2. Install dependencies:

    pip install -r requirements.txt

  3. Try out the example notebooks:

    • notebooks/linear_regression.ipynb
    • notebooks/classification_example.ipynb

You can open the notebooks in Jupyter or VS Code and run all cells to see ByteTorch in action.

Example notebooks

This project is always changing. If you find a bug or want to add something, go ahead. Have fun!